THE EFECTIVENESS OF TROPICAL CLOUD FOREST RESTORATION ON BIRD COMMUNITY AT CLOUDBRIDGE NATURE RESERVE, COSTA RICA MÉLANIE THIERRY, MAY 2016 KEY WORDS Reforestation, land-use change, forest restoration, bird community, landscape composition INTRODUCTION All around the world, forest and woodland vegetation have been removed and replaced by agricultural lands, leading to habitat loss and fragmentation (Foley et al. 2005), and decreased biodiversity (Fahrig 2003). In the tropics, more than 7 million hectares of forest were lost between 2000 and 2005, mainly because of land conversion from forest to pastures (FAO 2005). This phenomenon is still a major threat to Neotropical montane cloud forests (Brown et al. 2001). Cloud forests, characterized by a persistent, frequent or seasonal low-level cloud cover usually at the canopy level, represent a habitat of ecological importance for avian conservation, as they maintain a large number of endemic and threatened resident bird species, and provide wintering habitat for many migratory species. In particular, the highlands of Costa Rica and Panama are recognized as an Endemic Bird Area, which is an area where the distributions of two or more restricted-range bird species (breeding range smaller than 50,000 km²) overlap (BirdLife International 2003). Many forest bird species could not survive outside of forest in Costa Rica (Daily et al. 2001). It is therefore important to preserve and restore montane cloud forests in order to enhance biodiversity and protect endemic species associated with this unique habitat. Birds are considered good indicators of habitat change (Bradshaw et al. 2002). Along with biodiversity loss, land-use change has also caused shifts in the guild composition of bird communities. It has reduced the amount of frugivores, nectarivores and insectivores in favor of herbivores (Newbold et al. 2014). Bird species which are large-bodied, sedentary, forest specialists with long generation times, small ranges, small population sizes, and diets of fruit, nectar and invertebrates have been shown to be the most threatened by land-use change (Newbold et al. 2013 ; Barbaro and Van Halder 2009). Enhancing secondary growth permits the reconnection of forest fragments, allowing forest specialist birds to persist (Stouffer et al. 2006). Active restoration, by planting native tree seedlings, is expanding globally and contributes to the reduction of net loss of forest area in several countries (FAO 2005). This conservation strategy could allow the recovery of the Neotropical montane forest landscape matrix over time (Wijdeven and Kuzee 2000 ; Kappelle and Juárez 2006). In the tropics, the main limitation for secondary succession in abandoned pastures is seed availability 1 (Martínez-Garza and Howe 2003 ; Wijdeven and Kuzee 2000). Indeed, abandoned pastures lack seed banks of forest species (Wijdeven and Kuzee 2000). Animals (birds in particular), as pollinators, seed dispersers and consumers of herbivorous insects (Greenberg et al. 2000 ; Van Bael et al. 2007 ; Sekercioglu 2006), contribute greatly to forest regeneration. Planting tree seedlings could increase seed dispersal by animals (Lamb et al. 2005). It is therefore essential for the long-term success of forest restoration to restore bird communities as well, in order to reestablish the interactions between animals and their environment. Several studies found that restored forest can constitute habitat for birds, but the success of forest-dependent bird community recovery depends on the structure of the restored forest, plant species composition and age (Munro et al. 2011). Indeed, bird community composition depends on canopy cover, structure and species composition of the forest, which are linked to the tree species planted and the age of the forest. Ten years after forest restoration, reforested patches have been shown to recover about 50% of the species found in adjacent forest remnants (Smith et al. 2015). Munro et al. (2011) found that it takes about 30 years for restored forests to have similar bird communities as the adjacent forest remnants. But the mechanisms driving species assemblage observed at the patch scale might be influenced by factors at the landscape scale and the connectivity between patches (Fahrig 2003 ; Morrison et al. 2010). In fact, landscape effects might contribute more to species assemblage than do fragment size effects (Fahrig 2003). Some authors suggest that the relative importance of habitat factors versus landscape factors change depending on the stage of vegetation succession (Gould and Mackey 2015). However, to my knowledge, little is known about the mechanisms driving birds species assemblage after reforestation, especially at small scales such as the size of a nature reserve (Holl et al. 2000). This study takes place at Cloudbridge Nature Reserve in the Montane cloud forest in Costa Rica. Reforestation started in 2002 to restore forest on pasture lands. Nowadays, much of the reserve has been reforested by planting with tree seedlings of about fifteen native species. More than 270 bird species have been recorded within the reserve since 2002. In order to assess the effect of the forest restoration on the reserve’s bird community, a comparison of bird communities between reforested forest patches, natural regrowth patches, and primary forest remnants is needed. Bird species richness and diversity are expected to be higher at patches of intermediate succession stage than at younger patches and remnants, following the intermediate disturbance hypothesis (Connell 1978), as those patches can contain both forest-specialist and disturbance-tolerant species. If the reforestation of Cloudbridge Nature Reserve is successful, species and functional community composition of reforested patches is expected to become closer to the one of primary forest as the forest gets older, with more and more disturbance- sensitive species and less and less disturbance-tolerant species. In order to test these hypotheses, I conducted a bird survey between March and April 2016 to compare bird community species richness, diversity and composition between reforested patches, natural regrowth at different successional stages, and primary forest. 2 MATERIALS AND METHODS STUDY SITE Cloudbridge Nature Reserve is located in San Gerardo de Rivas, Perez Zeledon Province, Costa Rica. It borders Chirripo National Park to the east, a mosaic of agricultural land to the west, and relatively intact forested tracts of land both north and south. The reserve is a mosaic landscape of primary and secondary forests, riparian regions and tree plantations of different age covering about 400 hectares and ranging in elevation from 1550 and 2750 m. In this area, the average rainfall is approximately 4300 m per year and the mean low and high temperatures are 13.4 and 23.1 degrees Celsius. BIRD SURVEY FIGURE 1 : BIRD POINT COUNTS LOCATION AT CLOUDBRIDGE NATURAL RESERVE, SAN GERARDO DE RIVAS, PEREZ ZELEDON. The bird survey was conducted in March and April 2016 during the dry season using the point count technique, as bird data collected by this method can be directly related to habitat (Bibby 2000). To allow for safe and efficient access, all sites were located on existing trails. Twenty-four point counts, separated by a minimum distance of 200 m to ensure data independence (Bibby 2000), were distributed on 5 trails (Figure 1). Each trail was surveyed every week for three weeks per month (5 days per week). Bird point stations on each trail were surveyed from closest to furthest one week, and the opposite direction the following next week. Over 20 minutes, the same observer 3 recorded all the birds seen and identified at the species level in a radius of 25m, between the ground and the top of the canopy. Each bird station was marked with a fixed sign, which acted as the center of the study area. Only birds observed using the habitat within the study area were included (i.e. birds passing through the study area without stopping were not included). It was assumed that the probability of detection was uniform within this radius (Ralph and Scott 1981 ; Petit and Petit 2003). Each survey was done between 6am and 10am. LANDSCAPE VARIABLES FIGURE 2 : BIRD POINT COUNTS AND LAND COVER TYPE IN A 250M RADIUS AROUND THEM AT CLOUDBRIDGE NATURAL RESERVE. Land cover mapping was drawn with QGis 2.14.1 based on data collected at the reserve in 2007 and completed by aerial photo interpretation and field validations. Land cover was separated into 7 classes (habitat types): planted forest younger than 10 years, planted forest between 10 and 30 years, natural regrowth between 10 and 30 years, natural regrowth between 30 and 70 years, primary forest (older than 70 years), pastures, and habitation. No planted forest was older than 30 years and no naturally regenerated forest was younger than 10 years or older than 70 years. The percentage of each land cover class in a buffer of 250m around each bird point count (Figure 2) was used for analysis. The 250m buffer was considered the most adapted to the small reserve scale. For each point, the nearest distance from each point count to pastures was calculated. The altitude (from Google Earth) and the habitat 4 type (land cover class) in which the point was situated were also used for the analysis. Correlations between variables are presented in Appendix 1. BIRD TRAITS Species traits considered were diet, migratory status, IUCN status, forest specialization, body mass, generation length and range size (see Table 1 for traits description).